文献类型：期刊文献

英文题名：A novel TSM-RBF-driven hybrid detection paradigm for marine oil spills in shipborne radar images

作者：Xu, Jin[1,2];Xu, Bo[1];Zhang, Hao[3];Qian, Lihui[4];Yao, Boxi[1];Guo, Zekun[1];Wang, Haixia[5]

机构：[1]Guangdong Ocean Univ, Naval Architecture & Shipping Coll, Zhanjiang 524091, Peoples R China;[2]Guangdong Ocean Univ, Guangdong Prov Key Lab Intelligent Equipment South, Zhanjiang 524088, Peoples R China;[3]Natl Marine Environm Monitoring Ctr, Dalian 116023, Peoples R China;[4]Huainan Normal Univ, Sch Comp Sci, Huainan 232038, Peoples R China;[5]Dalian Maritime Univ, Dalian 116000, Peoples R China

年份：2027

卷号：333

外文期刊名：EXPERT SYSTEMS WITH APPLICATIONS

收录：SCI-EXPANDED(收录号：WOS:001864410300001)、、WOS

基金：This research was supported the Guangdong Basic and Applied Basic Research Foundation, grant numbers 2025A1515010886, 2023A1515011212, National Natural Science Foundation of China, grant number 52271359, the Special Projects in Key Fields of Ordinary Universities in Guangdong Province, grant number 2022ZDZX3005, the Shenzhen Science and Technology Program, grant number JCYJ20220530162200001, Postgraduate Education Innovation Project of Guangdong Ocean University, grant numbers 202421, 202539, 202551, the Guangdong Provincial Key Laboratory of Intelligent Equipment for South China Sea Marine Ranching, grant number 2023B1212030003, the Research Project of Higher Education Institutions in Anhui Province, China, grant number 2024AH051743.

语种：英文

外文关键词：Oil Spill; Marine radar; SBR; TSM-RBF

外文摘要：The continued expansion of global maritime transportation has led to a significant increase in the risk of offshore oil spill incidents, posing serious threats to marine environments and socioeconomic security. And low contrast between oil films and seawater backgrounds and strong clutter interference are prevalent in marine radar images. As a result, existing detection methods still exhibit notable deficiencies in target separability and boundary localization accuracy. To address the above issues, an optimized Topological Slime Mould-Radial Basis Function (TSM-RBF) hybrid oil spill detection algorithm was proposed. The SBR feature space was established from the preprocessed images. Local multidimensional feature values were evaluated using a sliding-window approach to enhance the brightness contrast between oil slicks and the surrounding seawater. An optimized TSM algorithm was subsequently introduced to perform global clustering optimization on the SBR feature map. The adaptive search model was constructed by fusing population-based search information to identify cluster centers that closely matched the pixel gray-level distribution in the gray-level space. The RBF neural network was further introduced to perform nonlinear mapping correction of boundary pixel classification results. Experimental results demonstrated that the proposed TSM-RBF hybrid algorithm outperformed comparative methods in sea surface oil film detection.